Corbell

Overview

Corbell is a comprehensive architecture intelligence tool that transforms codebases into queryable knowledge graphs. It offers MCP tools for service dependency queries, architecture context discovery, and semantic code search across multi-repo systems. The platform helps teams generate specifications that respect established patterns, review architectural decisions against actual code structure, and export task breakdowns to project management tools. All features run locally with no cloud dependencies, using SQLite for graph storage and supporting multiple LLM providers.

Try asking AI

After installing, here are 5 things you can ask your AI assistant:

• graph_query

  Query service dependencies, methods, and call paths

  Try:Find all services that depend on the auth-serviceTry:Show me the call path from user login to payment processing
• get_architecture_context

  Auto-discover relevant services for a feature description

  Try:What services are involved in processing a payment?Try:Find all services related to user authentication
• code_search

  Semantic search across the code embedding index

  Try:Search for payment processing retry logicTry:Find all implementations of rate limiting
• list_services

  List all services in the workspace graph

  Try:Show all services in this workspaceTry:List services with Go language

Comparable tools

Installation

pip install corbell
# With LLM support:
pip install "corbell[anthropic]"    # Claude
# Or for MCP:
corbell mcp serve

{
  "mcpServers": {
    "corbell": {
      "command": "corbell",
      "args": ["mcp", "serve"]
    }
  }
}

FAQ

What file systems does Corbell support?

Corbell works with any Git repository regardless of hosting (GitHub, GitLab, Bitbucket, etc.) and supports multiple programming languages including Python, JavaScript, TypeScript, Go, Java, C#, Rust, Ruby, and PHP.

How does Corbell ensure security?

All processing happens locally on your machine with no cloud dependencies. Code embeddings and architecture graphs are stored locally in SQLite. No code or context is sent to external servers unless you explicitly configure an LLM provider.

On Hacker News

Recent discussion from the developer community.

• Make your coding models create ADRs before implementation▲ 13💬 1
  Story by mercurialsolo · 2026-03-17
• Tool to reduce AI tokens and generate more detailed and accurate PRDs▲ 4💬 2
  Story by holy_pharaoh · 2026-03-27
• A tool to map dependencies across repos▲ 3
  Story by holy_pharaoh · 2026-03-26

Compare Corbell with